Gen Report Date: 2024-07-18
Years: 2022, 2023
Channels: XStore Retail
Measures: Product Revenue($)
Model: Add SUB_BRANDs’ Promotions as Regressors
RSquare: 0.7149872, MAPE: 0.6236067, MDAPE: 0.2895716
If removed the outliers (no. of outliers = 3) that Difference% > 100%, the performance as below:
RSquare: 0.8628255, MAPE: 0.256907, MDAPE: 0.2702861
These metrics are commonly used to evaluate the performance of a sales forecasting model.
R-squared, also known as the coefficient of determination, measures the proportion of variance in the dependent variable (in this case, sales) that is predictable from the independent variable(s).
Range: 0 to 1
Interpretation: 0.8628255 means that approximately 86.28% of the variance in the sales data can be explained by the model.
Assessment: This is a relatively high R-squared value, indicating that the model fits the data well. Generally, an R-squared above 0.7 is considered good for sales forecasting.
MAPE (Mean Absolute Percentage Error): 0.256907
MAPE measures the average of the absolute percentage differences between the forecasted values and the actual values.
Calculation: Mean(|Actual - Forecast| / Actual) * 100
Interpretation: On average, the forecast is off by about 25.69% from the actual values.
Assessment: While there’s no universal standard, a MAPE of 25.69% suggests moderate accuracy. For sales forecasting, this might be considered acceptable, but there’s room for improvement.
MDAPE (Median Absolute Percentage Error): 0.2702861
MDAPE is similar to MAPE but uses the median instead of the mean, making it less sensitive to extreme errors.
Overall Assessment:
The model shows good explanatory power (high R-squared) but moderate predictive accuracy (MAPE and MDAPE around 25-27%). This suggests that while the model captures the general trends in the data well, there’s still a notable margin of error in its predictions. Depending on the specific requirements of your sales forecasting application, this level of accuracy might be acceptable, but there could be room for improvement, especially in reducing the percentage errors.